Research map / Graph learning / Graph self-supervised learning
Adaptively Connected Neural Networks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace ResNet50 3×3 convolutions with adaptively connected ACNet layers | ResNet50 · ImageNet-1k val | top-1 accuracy: 76.4 -> 77.5 | Helped |
| Use generalized ACNet with fewer parameters | ImageNet-1k classifier · ImageNet-1k val | top-1 accuracy: 76.4 -> 76.2 | No clear effect |
| Replace CNN layers with ACNet layers | ResNet50-FPN Mask R-CNN · COCO val2017 | AP bbox: 38.0 -> 39.5 | Helped |
| Replace CNN layers with ACNet layers | ResNet50-FPN Mask R-CNN · COCO val2017 | AP mask: 34.6 -> 35.2 | Helped |
| Add ACNet | TriNet + Era. · CUHK03 | rank-1 accuracy: 62.0 -> 64.3 | Helped |
| Add ACNet to TriNet + Era. with reranking | TriNet + Era. + reranking (bs = 32) · CUHK03 | rank-1 accuracy: 61.2 -> 64.8 | Helped |
| Add global inference (ablation) | ACNet · CIFAR-10 test | error: 7.1 -> 6.0 | Helped |
| Add local inference (ablation) | ACNet · CIFAR-10 test | error: 24.0 -> 6.0 | Helped |
| Adaptively switch global and local inference instead of fixing their importance degrees (ablation) | ACNet · CIFAR-10 test | error: 6.8 -> 6.0 | Helped |
| Add global inference (ablation) | ACNet · Cora | accuracy: 82.1 -> 83.5 | Helped |
| Add local inference (ablation) | ACNet · Cora | accuracy: 76.3 -> 83.5 | Helped |
| Adaptively switch global and local inference instead of fixing their importance degrees (ablation) | ACNet · Cora | accuracy: 82.7 -> 83.5 | Helped |
Closest papers by meaning
- Adaptive Neural Networks for Efficient Inference (ICML 2017 · similarity 0.86)
- The Unreasonable Effectiveness of Fully-Connected Layers for Low-Data Regimes (NeurIPS 2022 · similarity 0.85)
- Approximation and non-parametric estimation of ResNet-type convolutional neural networks (ICML 2019 · similarity 0.84)
- Computational Separation Between Convolutional and Fully-Connected Networks (ICLR 2021 · similarity 0.84)
- Learning Transferable Architectures for Scalable Image Recognition (CVPR 2018 · similarity 0.83)
- PICNN: A Pathway towards Interpretable Convolutional Neural Networks (AAAI 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.